Penalizing Unfairness in Binary Classification
This addresses fairness issues in high-stakes domains like lending and admissions, but appears incremental as it builds on existing fairness criteria.
The paper tackles unfairness in binary classifiers by aiming to equalize false positive and false negative rates across two populations, and demonstrates its approach on datasets from criminal risk assessment, credit, lending, and college admissions.
We present a new approach for mitigating unfairness in learned classifiers. In particular, we focus on binary classification tasks over individuals from two populations, where, as our criterion for fairness, we wish to achieve similar false positive rates in both populations, and similar false negative rates in both populations. As a proof of concept, we implement our approach and empirically evaluate its ability to achieve both fairness and accuracy, using datasets from the fields of criminal risk assessment, credit, lending, and college admissions.